4 points | by tattattaei 9 hours ago ago
2 comments
It may be a positive feedback loop - whichever language the AIs write more of becomes more predominant in future training data.
But if AIs write bugs in that language, they also poison the well...
Model collapse of the winning language
It may be a positive feedback loop - whichever language the AIs write more of becomes more predominant in future training data.
But if AIs write bugs in that language, they also poison the well...
Model collapse of the winning language